Transform functions should bypass functionConfigs

[prachirp]

We don't want functions to manipulate the function config if it's in the same directory as KRM resources.

We need a way to bypass this.

Pain point example: kptdev/krm-functions-catalog#87

[mortent]

@prachirp So this the idea here that resources that have the config.k8s.io/function annotation should not be mutated by a function pipeline? There are scenarios where we want to put the config annotation on normal resources, so I'm not sure if that would work. This seems very similar to #630.
Another option might be that for resource that are only present to declaratively define a KRM function should not be of the ConfigMap type, but instead we could use a client-side CR where the Kind would be Function.

[prachirp]

@mortent Yes, that's a good summary. And I agree #630 is similar but not the same. I'm not sure what the solution should be here. It seems just defining a new annotation is a much simpler fix than using a new Function kind for function configs.

From Frank: #630 is a different issue, it's about the inventory object being processed. Function configs are different. Let's file a separate bug for this.